the plate virulence test diphtheriafailed to give a halo, and 3.4% ofavirulent strains gave false...

J. clin. Path. (1949), 2, 250.

THE PLATE VIRULENCE TEST FORDIPHTHERIA

BY

STEPHEN D. ELEK

From the Department of Bacteriology, St. George's HospitalMedical School, London*

(RECEIVED FOR PUBLICATION, MAY 20, 1949)

A method has recently been described (Elek,1948) for the detection in vitro of toxin-producingorganisms including C. diphtheriae. The presentpaper has two purposes. The first is the examina-tion of a larger series of strains of C. diphtheriaewith a view to establishing the reliability of theprocedure, for if the method is to be of anypractical value for clinical purposes it mustdemonstrate all toxin-producing strains within areasonably short time. The second purpose is thestandardization of the ingredients used so as toyield the best results and to ensure that the condi-tions of the test may be accurately reproducible indifferent laboratories.

Petrie and Steabben (1943) first suggested theuse of nutrient media as a matrix for toxin-antitoxin reactions. They evolved a method forthe detection of toxicogenic clostridia and brieflyreported some experiments with diphtheria bacilli.Ouchterlony (1948) showed that when diphtheriaantitoxin was incorporated in a serum ditch plateand a toxin-producing strain streaked across,several lines were produced, one of which wascaused by the toxin. For the practical test of hisin vitro method he incorporated various concentra-tions of antitoxin in serum agar plates, as didPetrie and Steabben (1943), and each strain wastested on a series of plates with falling antitoxinconcentrations. The appearance of a halo aroundthe inoculum was regarded as positive. Two seriesof strains were examined by this method and bythe subcutaneous guinea-pig test. In the first seriesof 237 strains there was a discrepancy amountingto 8.5%. Of the guinea-pig positive strains 5.1%failed to give a halo, and 3.4% of avirulent strainsgave false positives. In the second series of 308strains the discrepancy was 6.2%, and included4.2% false positives. Elek (1948) criticized Petrie

*Part of Ph.D. (London) thesis submitted July, 1948.

and Steabben's method on the grounds that a ringin a serum plate does not necessarily signify atoxin-antitoxin reaction and described a new tech-nique using a medium better suited for toxin pro-duction. The antitoxin gradient was set up bymeans of a filter paper strip dipped into highly con-centrated refined antitoxin. This system was foundto be free from the zone effect and the theoreticalconsiderations are dealt with in a later paper (Elek,in press). In a small series complete agreementwas found between the in vivo and in vitromethods, and furthermore the results could beread within 24 to 48 hours compared with the48 to 96 hours required by the method using serialplates containing graded quantities of antitoxin.Carter and Wilson (1949), using Elek's method butsubstituting human serum for horse serum, foundabsolute agreement with 200 strains between thein vivo and in vitro tests. Ouchterlony (1949) ina later paper reports the use of serum ditch plates.Readings were taken at 24-hour intervals for up tofour days.Concerning the constituents of a medium

designed to give a good yield of toxin muchinformation is already available. Almost all thework however was carried out with the classicalPark-Williams No. 8 strain, and it is by no meanscertain that a medium giving optimal toxin produc-tion with it will necessarily meet the requirementsof all other toxin-producing strains. Until about25 years ago it was widely believed that only meatinfusion media could be used for the culture ofdiphtheria bacilli. Davis and Ferry (1919) statedthat it was impossible to obtain growth unless0.2% meat infusion was present, and 10% wasneeded for toxin production. Wadsworth andWheeler (1928) succeeded in producing toxin onpeptone without meat infusion, but with variouschemicals added. The addition of glucose to

copyright. on F

ebruary 28, 2020 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.2.4.250 on 1 N

ovember 1949. D

ownloaded from

http://jcp.bmj.com/

THE PLATE VIRULENCE TEST FOR DIPHTHERIA

infusion media as a source of energy was suggestedby various authors (Park and Williams, 1896;Smith, 1899; Locke and Main, 1928; Ramon,1929; Hazen and Heller, 1931) and sodium lactatewas introduced by Wadsworth and Wheeler in1928. Pope (1932a) in a very careful reinvestiga-tion of the effect of various carbohydrates andorganic acids found that a concentration of 0.6%lactic acid was optimal. He used proteose-peptonefor his media and the Park-Williams No. 8 strainfor toxin produ-tion. He concluded that acclimat-ization to the medium was unimportant, but severeheat treatment, such as autoclaving of the medium,would destroy toxin production without interfer-ing with growth. Pope and Healey (1933a and b)examined maltose concentrations and found 0.4%to be the optimum, 0.8% yielding decidedly lesstoxin. They found further that the best growthof the organism did not necessarily yield the maxi-mum amount of toxin, but that this varied with themedium. They also established the very importantfact that the initial presence of the toxin was with-out effect on growth and further toxin production.Finally they proved that toxin production wasmaximal, other things being equal, when growthoccurred at the air/liquid surface. The results ofthese observations were incorporated in themedium previously described (Elek, 1948).

Experimental Work and MaterialsSubsequent work has shown that successive batches

of the medium showed some variation, and slight

G = grM mlI = iniI = R=Re

48

c)

0I

z2 24

z~~

1 2 3 4 5 6G M I M G G

DIFCO PROTEOSE-PEPTONE

modifications have been made. Charcoal-clearing hasbeen omitted, as filtering through paper pulp resultedin a sufficiently clear medium. The method of adjust-ing the reaction had to be standardized as this stagedetermines the precipitation of phosphates, and conse-quently the iron content of the medium. The prepara-tion of the medium was accordingly modified asfollows.Twenty grammes of proteose-peptone (" difco "), 3 g.

of maltose, and 0.7 ml. of lactic acid (B.P.) were dis-solved in 500 ml. of distilled water. To this 1.5 ml.of 40% NaOH was added and the medium well shakenand heated to boiling point. The deposit was filteredoff through filter paper and the reaction adjusted topH 7.8 by adding normal HCI. Then 500 ml. of 3%agar (powder) were prepared in I% NaCl and its re-action adjusted to pH 7.8. This was filtered throughpaper pulp in a Buchner funnel and then added to thefluid base. After distribution in 10 ml. quantities itwas sterilized by autoclaving for 10 minutes at 10 lb.To ensure an even thickness of the medium special

compressed flint-glass petri dishes, 32 in. in diameter,were used. As these are mould-compressed thebottoms are absolutely flat, and this combined witha standard amount of medium for pouring (12 ml.)gives an even thickness in all the plates. The plateswere poured on a levelling board supported on threepoints and set horizontal with a spirit level. Forroutine purposes neither the mould-compressed platesnor the levelling board is absolutely necessary. Com-mercial refined diphtheria antitoxin globulins wereused diluted with sterile carbol-saline (0.5% phenol innormal saline) to a concentration of 1,000 units perml. This preparation of antitoxin contains 0.35% ofcresol, but as this does not diffuse out it does not

vis interfere with the reaction. Inis fact dilution with carbol-saline isSi useful as it reduces the risk of!rmedius contamination. A filter stripiults with batch A measuring 60 mm. by 15 mm.ults with batch B sterilized by dry heat was

moistened with the antitoxin byimmersing it completely or drop-ping the antitoxin on to it. Plateswere poured with 10 ml. of themedium and 2 ml. of normalhorse serum and the filter stripwas embedded in the still fluidmedium leaving an unbroken sur-face for inoculation. Two brandsof maltose were tried. One wasan old batch without a manufac-turer's name and the other was"difco." They gave identical re-

FIG. 1.-Comparison, on the basis7 B of the time required for pos-I 4~itive results, of two batches

of medium (strains 1 to 8).

251

copyright. on F


http://jcp.bmj.com

/J C


ovember 1949. D

ownloaded from

http://jcp.bmj.com/

STEPHEN D. ELEK

sults. Two samples of lactic acid were tried, andalthough both complied with B.P. requirements, onlyone gave satisfactory results, no lines being producedwith the other within 48 hours.Method for Assessing Media.-It was found that

different batches thus prepared did not show appreci-able variations. Furthermore, the time of appearanceof the toxin-antitoxin line for a given organism ondifferent batches was roughly the same. Fig. 1 showsa set of experiments illustrating this point. Twoidentical batches of media were prepared withproteose-peptone (" difco ") and the same batch ofhorse serum was used for enrichment of both. Eightvirulent strains of C. diphtheriae were used consist-ing of three gravis, three mitis, and two intermediusstrains. The tests were carried out in the ordinaryfashion by streaking each organism at right angles tothe filter strip. The width of the streak was notstandardized beyond using the same loop. Readingswere taken at three-hourly intervals, day and night,for 48 hours and the earliest appearance of the linesrecorded. The maximum time difference for any onestrain on the two media was about 15 hours, and theaverage about fivehours. Clearly thetime of appearance of DAYSthe lines can serve asa rough measure of 5

the suitability of themedium for toxin pro-duction. In this waya comparison can bemade, for instance,between the efficacy, 4for the purposes ofthis test, of variouscommercial peptones. <

Choice of Peptone.-It is well knownthat media preparedwith different commer-cial peptones showgreat variation intoxin production.Thus Hosoya, Ozawa,and Tanaka (1933)found that Chapoteaupeptone was highly

FIG. 2.-Effect ofvarious peptoneson toxin produc-tion. Suitabilityofthe peptone forthe purposes ofthe test is inverse-ly related to theheight of thelines. Strains I

to 8 identical withthose in Fig. 1.

A.4

03m

z0

I~-

U

w

0:

U. 2.

cc4

~0

w

Uw

z

z

ACto

U

Z

i 2345S 6

DiFCO

potent, yielding a high lethal dose and Lf 10.5,whereas the same medium prepared with Witte'speptone or Teruuchi peptone gave toxins of lowpotency only. With "difco" proteose-peptoneAndo and Komiyama (1935) obtained yields of Lf 18or more, thus confirming Pope's results. Apart from" difco " proteose-peptone (Fig. 1), five further brandsof peptone were tested by substituting them in thepreparation of the medium described. These were" difco " peptone, " eupeptone No. 1," Evans pep-tone, "oxoid," and Witte's peptone. All the mediawere prepared in exactly the same manner and thesame eight strains were used for all these experiments.The same batch of serum (serum A in Fig. 3) was usedthroughout. The plates were incubated for five days,readings being taken three-hourly day and night forthe first 60 hours, and thereafter every 12 hours.Fig. 2 shows the results obtained, the height of thelines representing the incubation necessary to obtaina positive iesult. An. arrow indicates absence of lineproduction within the five days of the experiment. Itwill be seen that two of the peptones, "difco" andEvans, gave positives with all the eight strains well

I I

.1*

III I I I II I==I I I IIiI rn-i-r= * mu * .1240 123450I-. ....I 22A9

1 23P 4 5 67

EU PEPTONE£1 2 3 4 5 6

EVANS3O4 6 7OXOID

4 .,

WI TTE

'1

2)52

a a

copyright. on F


http://jcp.bmj.com

/J C


ovember 1949. D

ownloaded from

http://jcp.bmj.com/


within 48 hours. The general impression one gainsis that neither is quite as potent as the "difco"proteose-peptone (compare Fig. 1). Evans peptone,however, is almost as good, and " difco " peptonecomes third. " Eupeptone No. 1," "oxoid" andWitte's peptone are clearly unsuitable for the pur-poses of this test. Witte's peptone gave positives withonly two strains within 48 hours (both gravis). Theother two peptones gave no positives in two days, thetime within which all virulent strains gave positiveresults with proteose-peptone. Evans peptone is prob-ably suitable for the test, but testing with a muchlarger series is required. So far this has been doneonly with " difco " proteose-peptone.

Iron and Copper Content.-Locke and Main (1930)found that media usually used for toxin productioncontained about 1-4 mg. of copper per litre, and about0.1 to 0.4 mg. of iron per litre. Pope (1932b) re-investigating the problem found that the addition ofcopper to media had little effect as long as it did notexceed 8 to 10 mg. per litre, which was the maximumtolerated. He also found that filtered medium wasimproved by iron, but media sterilized by heat did notrequire additional iron. Amounts of 0.5 mg. per litrewere found to be definitely detrimental. The effect ofiron on toxin production has been further investigatedin recent years (Pappenheimer and Johnson, 1936;Mueller, 1941; Mueller and Miller, 1941). Theseworkers found that the optimal amount of iron variedgreatly with the composition of the medium and withthe strain. No attempt was made in the preparationof the media used in the present experiments to con-trol either the copper or the iron content, except bythe use of glass vessels and the careful control of theprecipitation of the phosphates. The iron contents ofthe broths prepared with various peptones were deter-mined in case the reason for the divergent results layin this. Table I shows the total iron content of thefluid base prepared with the six peptones tested.

TABLE IIRON CONTENT OF BROTH MEDIA PREPARED WITH

DIFFERENT PEPTONES

Peptone Fe (mg./100 ml.)

"Difco" proteose-peptone 0.021Evans ... 0.035"'Difco" ... 0.086Witte's ... 0.050"Oxoid" ... 0.046"Eupeptone No. 1" 0.042

Except for the addition of agar these media wereidentical with the ones referred to above. Clearlythe iron content does not explain the variation intoxin production although the best results wereobtained with the two peptones showing the lowestiron values. Considerations of the iron contentof the dry reagents (Table II) shows that little

TABLE IIIRON CONTENT OF DRY REAGENTS

Reagent Fe (mg./100 ml.)

Agar powder ... 5.64"Difco" proteose-peptone 3.93Evans peptone ... 4.35

significance can be attached to small variationsoccurring in the final medium. It can be calculatedthat 1.5% agar corresponds to 0.084 mg. Fe. per100 ml. For purposes of comparison this was leftout of account partly because it represents a constantaddition to all the media and partly as not all of it isin an ionizable form. The calculated amount of ironin 2% proteose-peptone is 0.078 mg. per 100 ml. asagainst the 0.021 mg. actually found. Thus, in thepreparation of the medium approximately three-quarters of the iron is lost with the precipitation ofthe phosphates. A similar loss can be calculated inthe case of Evans peptone. The essential fact is thatmedia prepared in glass vessels will yield satisfactoryresults without an adjustment or even determination ofthe iron content so long as the precipitation of thephosphates is carried out in a standard manner.Serum Enrichment.-The remaining variable is the

added sterile normal horse serum. Early in the courseof these experiments it was noted that the addition ofhorse serum improved the reaction. This effect maywell be due to enrichment of the medium leading toincreased toxin production, or the adjuvant effect maybe upon the flocculation, or possibly both thesecauses may be operative. It was assumed that eventhough the antitoxin in the strip might contain all thenecessary non-specific constituents for flocculation itwould be unlikely that they would follow the sameconcentration gradient as the antibody globulins. Onthe other hand the addition of normal serum to themedium will yield an even concentration of any non-specific substances that may be required to intensifythe flocculation lines. The views as regards the needfor non-specific substances in flocculation reactionsare somewhat contradictory, and the available infor-mation concerning their role and nature is insufficientfor application to this test in a systematic way. It wasdecided, therefore, to examine the problem empirically.

Although, as has been shown, an exact reproduc-tion of the base medium is relatively easily obtained,the addition of different batches of horse sera maymaterially affect results. This is clearly a consider-able disadvantage for a routine test, since in theabsence of knowledge of the various factors whichdetermine the usefulness of the horse serum little canbe done to obtain uniformity. Attempts to substitutebovine albumin for the horse serum were not success-ful. The sera used in the course of these experimentswere commercially obtained No. 2 normal horseserum, heat treated, and issued for the preparation of

253

copyright. on F


http://jcp.bmj.com

/J C


ovember 1949. D

ownloaded from

http://jcp.bmj.com/

STEPHEN D. ELEK

23456 78S. 12 34 5678 1 2 3 456 738 123 4 5678 912 3 4 56 78 1 123456 7 8

SERUM A SERUM B SERUMi C UNIT A S UNITS A 10 UNITS A

ANTITOXIN ANTITOXIN ANTITOXIN

FIG. 3.-Effect of different batches of horse serum (A, B, and C) and the effect of adding diphtheria antitoxin to

serum A. Base medium prepared with " difco " proteose-peptone. Strains 1 to 8 same as in Figs. 1 and 2.

media by the Welicome Physiological Research

Laboratories. The results obtained with three

different batches of horse sera are shown in Fig. 3.

Two of these (A and B) were No. 2 Wellcome brand

sera and the third (C) was a normal horse serum

obtained from a different source. The same eight

strains of diphtheria bacilli were used as in the

previous experiment. The base medium was the same

batch for all, made with proteose-peptone. Readings

were taken as before. Sera A and B did not differ in

any material way, and both yielded positives for all

the strains well within 48 hours. Serum C, on the

other hand, required longer incubation for the lines to

develop, and two strains failed to give positive results

within 48 hours, the lines becoming visible a few hours

later. Clearly with this serum one would miss two

virulent strains unless readings were taken after two

days. It is hardly possible to conjecture why this

serum was inferior to the others, but one possible

explanation might be that it containe'd natural diph-

theria antitoxin which interfered with the reaction.

Normal horse sera of course vary in this respect, and

in some unselected samples the titre may be quite

appreciable. It seemed interesting to test the hypo-

thesis whether a"bad " serum could be unsatisfactory

on account of a high antitoxin level. To serum A

varying amounts of crude horse antitoxin were added

to yield concentrations of unit, 5 units, and 10 units

254

DAYS

41wI.4c

0

z0

0

U,Sdu

0I.

z

copyright. on F


http://jcp.bmj.com

/J C


ovember 1949. D

ownloaded from

http://jcp.bmj.com/


per plate. As the volume per plate was 12 ml. thiscorresponds to 0.08 unit, 0.42, and 0.83 units respec-

tively per ml. of medium. Fig. 3 shows the effect ofthe added antitoxin: with 1 unit per plate five strains-ave positive results within 48 hours, but three strains,ne gravis and two intermedius, failed to produceaes even in five days. Five units per plate almostompletely inhibited the reaction, but two strains justnanaged to produce lines within five days. It is clear'rom this that a high natural antitoxin level will give-n unsatisfactory serum for this test. On the otherhand there is no evidence that this is the only reason

'or a serum proving unsatisfactory. It was found thatthe addition to serum C of purified diphtheria toxoidin amounts sufficient to over-neutralize any naturalantitoxin did not improve it. This variation in the serumis at present difficult to overcome and represents themost serious disadvantage to routine use of thismedium. Admittedly only small quantities of serum

are required per plate, and each plate can be used forabout six strains. Furthermore, a batch of serum

once tested and found satisfactory can be kept in therefrigerator for long periods without apparent deterior-ation. The difficulty was the initial testing of thebatch. At first a known strong toxin-producing strainwas included on every plate, but this is obviously not

satisfactory since weak toxin-producers can be over-

looked in this way. Dr. A. T. Glenny suggested theuse of a strain which he used for many years as a

control in intracutaneous virulence tests. This strain,which he kindly let me have, was maintained in thefreeze-dried state and has a fixed low virulence: inthe ordinary skin tests it gives a reaction which can

just be accepted as positive. On testing with the platemethod it was regularly late in producing a line, andwith serum A it required about 40 hours' incubationfor a positive reaction to appear. Two further strainswere encountered which showed similar behaviour,one a gravis and the other an intermedius strain. Bothof these were reported as giving a very weak intra-cutaneous virulence test, and this had its counterpartin a weak line appearing only towards the end of thesecond day's incubation. At present this biologicalapproach constitutes the only way of standardizing theserum. Once a batch is found satifactory it can bekept at low temperature, but it is advisable to includeone of the weakly toxicogenic strains with each seriesof tests as a control.

The Reading of the Test.-The reading of the testpresents no difficulties and requires no previous experi-ence. The lines at their earliest appearance requireoblique illumination and a hand lens. Good obliqueillumination can be obtained by mounting two pieces ofphotographic paper between half-plate glass leaving astrip of about 1 cm. in the middle. This slit is heldbehind and parallel to the filter strip. By moving itslightly up or down in front of an electric bulb, a veryearly reading of the lines can be taken. However, a

few hours later they become readily visible even bytransmitted light. By the end of 24 hours the largemajority of the virulent strains showed lines easilyvisible by transmitted light.

Comparison of In Vivo and In Vitro TestsThe final assessment of the method from the practi-

cal point of view lies in its usefulness when applied toa larger number of strains of C. diphtheriae and in itsagreement with the accepted animal tests of virulence.When we examine the tests used as a routine forestablishing whether a strain is " virulent " or not, wefind that the test, subcutaneous or intradermal, isbased' on specific neutralization of toxin produced invivo. The difference between the established viru-lence tests for diphtheria bacilli and the in vitro testdescribed here, is only the difference between theconditions leading to toxin production. In otherwords, the question to be answered is whether it ispossible to produce a nutrient medium catering forthe individual requirements for toxin production of allthe strains encountered and whether the test is sensi-tive enough to detect the presence of the toxin thusproduced. The results of parallel tests carried out on

135 strains are shown in Table III, the same strainsgiving identical results by each method. These strainswere mostly isolated and tested during the latter partof 1947 in London. The determinations of type andguinea-pig virulence were carried out independentlyof the plate tests by the bacteriologists who suppliedthe strains. Of the 67 strains described as gravis onlyone was avirulent, and this was obtained from theNational Collection of Type Cultures. Of the 45mitis strains, on the. other hand, no less than 13 were

avirulent, and all but one of these were isolated in thecourse of routine work. Of the 23 intermedius anduntyped strains three were avirulent.

TABLE IIICOMPARISON OF GUINEA-PIG AND PLATE VIRULENCE TESTS

Guinea-pig Test Plate Test Total No.Type of C. diphtheriae Positive in Positive in Negative in

Virulent Avirulent 24 Hours 48 Hours 58 Hours

Gratis .66 1 61 66 1 67

Mitis 32 13 27 32 13 45

Intermedius and untyped 20 3 14 20 3 23

255

copyright. on F


http://jcp.bmj.com

/J C


ovember 1949. D

ownloaded from

http://jcp.bmj.com/

STEPHEN D. ELEK

DiscussionThe fact that the agreement in this series was

absolute suggests that the sensitivity of the twotests is of a comparable order, and this is furtherborne out by the findings with the weakly toxi-cogenic strains already referred to. This is sur-prising indeed as the amount of diphtheria toxinrequired to produce a positive skin test in aguinea-pig is extremely minute, whereas the mini-mum amount of toxin required to produce a visibleline in the plate is considerable and amounts toabout 6 units. The two tests are, however, notcomparable on the basis of the quantity of toxinalone since the conditions under which the organ-isms grow are vastly different. In the plate tech-nique the organisms grow unhindered and asurplus of nutrient material designed for toxinproduction is available so that the accumulation oftoxin is continuous.From these observations it would seem natural

to connect the time of appearance of the lineswith the amount of toxin produced. Clinically andepidemiologically a method giving quantitativeinformation of an organism's toxin-producingcapacity would be of value. In model experimentsusing one filter strip soaked in toxin and the otherin antitoxin, the position and angle of the line wererelated to the quantities used. With a streak ofinoculum, however, visible growth precedes for avariable time the formation of the toxin-antitoxinline. The position of the line formed is dependenttherefore on the actual shape of the antitoxingradient at the time when toxin begins to be pro-duced in an appreciable concentration. It is notpracticable, for routine purposes, to measure thedistance of the line from the filter-strip. On theother hand it was found that a given strain will pro-duce a line within more or less the same time onrepeated examinations, and it seems that the timecan be taken to reflect a biological characteristic ofthe strain. Thus, although it is strictly speakingnot a measure of the amount of toxin formed bya given strain, it still yields information of interestconcerning the toxin-producing capacity.

Twenty-nine of the 135 strains tested wereobtained from the National Collection of TypeCultures. Of these 27 were found to be virulent:20 yielded positive plate tests within 24 hours andseven (26%) only after 48 hours' incubation. Ofthe remaining 106 strains, the majority of whichwere freshly isolated, 91 were virulent. Eighty-three of these showed a positive plate reactionwithin 24 hours, but eight strains required 48 hours'incubation for the positive reaction to appear. Thisconfirmed the impression that old laboratory

strains require rather longer incubation. The per-centage of strains requiring as long as 48 hours'incubation to produce a positive reaction was 26%of the old laboratory strains, but only 8.8% of thefreshly isolated virulent strains. Admittedly thenumbers in each group are small. It is alsointeresting to note that all the freshly isolatedgravis strains were virulent, and of these 27 werepositive in 24 hours, leaving only two strains thatneeded two days' incubation. In the group ofintermedius and untyped strains six were obtainedfrom the National Collection of Type Cultures,but a further three were stock cultures. Of theremaining 14 freshly isolated strains 12 were viru-lent and two were not. Of the 12 virulent strains10 gave positive results in 24 hours and the othertwo in 48 hours. Thus of the 88 freshly isolatedvirulent strains 81 (over 92%) were recognizableas such by means of the plate technique by the endof the first day. It was also observed that not onlyold stock cultures, but also freshly isolated strainscan give a sluggish reaction if the streak on thetest medium is made from a culture more than24 hours old. It is essential, therefore, to use ayoung culture. The original culture may beon inspissated serum, blood agar, or telluritemedium.

It can be concluded from these data that theroutine virulence test as performed by the intra-dermal method, and the in vitro test here described,give complete agreement as regards the 135 strainstested. This statement does not, however, invali-date the argument that true " virulence " of diph-theria bacilli is not synonymous with toxicogeni-city. If there is a difference and toxin productionis only one of the criteria of virulence, then wehave at present no method available to assess thesum total of the various factors entering into theconcept of virulence. It is clear, however, thattoxin producion is a sine qua non of virulence.It is not possible to say whether a reversion totoxin production occurs in the case of avirulentstrains. An interesting observation bearing on thisquestion was made in the case of an old typeculture strain (N.C.T.C. 322/Centry). This wasreceived as avirulent, having been tested in 1920by the bacteriologist who isolated it. It was thenfound that 4 ml. of a culture filtrate (further detailsnot available) were non-lethal to guinea-pigs.This organism was found to be toxicogenic in vitroand this was confirmed by an intradermal animaltest. Unfortunately the data of the original viru-lence testing were not sufficiently detailed and it isnot possible to say whether a reversion to virulenceoccurred in this case.

256

copyright. on F


http://jcp.bmj.com

/J C


ovember 1949. D

ownloaded from

http://jcp.bmj.com/


FIG. 4.-Looping of the toxin-antitoxin lines and thedemonstration of non-specific turbidities. Strains 1, 2,3, and 7 virulent in vivo (+). Strains 4, 5, 6, 8, and 9avirulent (-).

FIG. 5.-Plate photographed after prolonged incubationand several days at room temperature showing second-ary lines. Strains 1 and 3 are virulent, 2 is avirulent.Strain I shows two fine lines developing between thetoxin line and the filter strip.'T

The rule of looping can be made use of whendetermining minimal toxin production by a strain.Fig. 4 shows a plate with nine strains of whichNos. 1, 2, 3, and 7 were virulent (+) and Nos., 4,5, 6, 8, and 9 avirulent (-) in vivo. Strain No. 2is weakly toxicogenic and the lines produced bystrains No. 1 and No. 3 show marked bending onapproaching it. When the lines approach a strainthat is devoid of toxin production they remainstraight and cut the non-toxic streak ; that is, theline of strain No. 3 cuts across strain No. 4 whichis non-virulent. It should be stressed, however,that the looping effect only applies when bothlines are simultaneously produced. If one lagsbehind an angular joining of the lines occurs, asseen between strains No. 2 and No. 3, and this isdue to the rearrangement of the antitoxin gradientaround the line first formed. Any turbidity aroundthe streak should be disregarded, since manyorganisms produce non-specific turbidities inserum agar media. Thus strain No. 7, which isvirulent, produces a halo, but so does strain No. 8as well although it is avirulent.When the incubation is prolonged beyond 48

hours, or the plates left at room temperature forseveral days, very fine lines showing the typicalarrowhead form may appear (Fig. 5, strain No. 1).These lines are definitely not due to toxin produc-tion as they appear with both virulent and aviru-lent strains. To avoid confusion the plate testmust not be read later than 48 hours. The charac-teristics of these secondary lines are as follows:they appear late, usually after several days, theyare fine, and may number two or three. Apparentlythey are due to diffusible antigens produced bydiphtheria bacilli, the exact nature of whichrequires further elucidation.

In view of the secondary lines it is mostimportant that the medium used in the platetest should be capable of demonstrating minimalamounts of toxin production in 48 hours, that is,before the appearance of secondary lines mightlead to confusion. The medium, the prepara-tion of which is here described, satisfies thisrequirement.

Summary and ConclusionsA method is described for assessing the value of

various constituents in the preparation of mediafor the diphtheria plate virulence test.

Six different brands of peptone were found tovary in their suitability for the test.The iron content of the media required no

adjustment provided the preparation followed theroutine described.

257

copyright. on F


http://jcp.bmj.com

/J C


ovember 1949. D

ownloaded from

http://jcp.bmj.com/

STEPHEN D. ELEK

Batches of horse sera vary in their suitabilityfor the test and the value of the serum used hasto be established empirically.A series of 135 strains gave identical results by

the plate test and the intradermal guinea-pig test.The sensitivities of the two tests were found to beof the same order.Over 92% of freshly isolated virulent strains

gave a positive reaction within 24 hours with theplate test, and 100% were positive after 48 hours'incubation. Readings must not be taken later than48 hours.

I wish to express my sincere thanks to Dr. A. G.Signy, Dr. E. Straker, Dr. A. Beck, and Mr. H. Proomfor supplying various strains; to Dr. J. E. McCartneyand Mr. J. C. Monckton for virulence-testing as wellas supplying strains; to Professor T. Crawford forvaluable help in the preparation of the paper, andto Mr. T. Shaw and Mr. T. Pringle for technicalassistance.

REFERENCESAndo, K., and Komiyama, T. (1935). J. Immunol., 28, 345.Carter, H. S., and Wilson, W. (1949). Glasg. med. J., 30, 43.Davis, L., and Ferry, N. S. (1919). J. Bact., 4, 217.Elek, S. D. (1948). Brit. med. J., 1, 493.Hazen, E. L., and Heller, G. (1931). Proc. Soc. exp. Biol. N.Y., 28,

423.Hosoya, S., Ozawa, E., and Tanaka, T. (1933). Jap. J. exp. Med., 11,

463.Locke, A., and Main, E. R. (1928). J. infect. Dis., 43, 41.Locke, A., and Main, E. R. (1930). J. inject. Dis., 46, 393.Mueller, J. H. (1941). J. Immunol., 42, 343.Mueller, J. H., and Miller, P. A. (1941). J. Immunol., 40, 21.Ouchterlony, 0. (1948). Acta path. microbiol. scand., 25, 186.Ouchterlony, 0. (1939). Lancet, 1, 346.Park, W. H., and Williams, A. W. (1896). J. exp. Med., 1, 164.Pappenheimer, A. M., Jr., and Johnson, S. J. (1936). Brit. J. exp.

Path., 17, 335.Petrie, G. F., and Steabben, D. (1943). Brit. nmed. J., 1, 377.Pope, C. G. (1932a). Brit. J. exp. Path., 13, 207.Pope, C. G. (1932b). Brit. J. exp. Path., 13, 218.Pope, C. G., and Healey, M. (1933a). Brit. J. exp. Path., 14, 77.Pope, C. G., and Healey, M. (1933b). Brit. J. exp. Path., 14, 87.Ramon, G. (1929). C.R. Acad. Sci., Paris, 18s, 718.Smith, T. (1899). J. exp. Med., 4, 373.Wadsworth, A.,'and Wheeler, M. WV. (1928.) J. intject. Dis., 42, 179.

258

copyright. on F


http://jcp.bmj.com

/J C


ovember 1949. D

ownloaded from

http://jcp.bmj.com/

the plate virulence test diphtheriafailed to give a halo, and 3.4% ofavirulent strains gave false...

Documents